Structuring Hybrid Model Using Bilateral and Guided Filters for Image Denoising
Rama Lakshmi Gali, Manne Praveena, Santhosh Kumar Veeramalla · 2023
Image denoising is a foundational challenge within the realm of image processing. This study introduces an innovative technique for image denoising, leveraging a hybrid filtering method. The hybrid filter represents a fusion of two well-established filters, namely the bilateral filter and the guided filter. By amalgamating the merits of these filters, the proposed approach adeptly retains image edges while simultaneously mitigating noise. During the hybrid filtering procedure, a guided filter is employed to generate a guiding image that aids in edge preservation while smoothing the image. Subsequently, a non-linear filter known as the bilateral filter is applied to each pixel, its operation based on the weighted average of neighboring pixel intensity values, calculated through a Gaussian distribution. Experimental outcomes demonstrate that the proposed hybrid filtering method surpasses other state-of-the-art denoising techniques concerning peak signal-to-noise ratio (PSNR). The PSNR values achieved through the proposed method consistently range from 31 to 33 dB across various noise types present in the image. Consequently, the resulting images exhibit markedly reduced noise levels while sustaining their clarity. In summary, the proposed hybrid filtering method proves to be an effective strategy for image denoising, capitalizing on the strengths of both the bilateral and guided filters. This approach offers a pragmatic solution for real-world applications within the domain of image processing.